Models › Qwen3.5-9B

Qwen3.5-9B: VRAM needs and speed by GPU

At Q4_K_M, Qwen3.5-9B needs about 6.7 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.

8.95B parameters · license: apache-2.0 · 262,144-token context · GGUF published Feb 28, 2026 · GGUF files on Hugging Face

Memory needed per quantization

QuantFile sizeNeeds (estimate)Smallest card it fits
UD-IQ2_XXS3.2 GB4.2 GB8 GB cards
UD-IQ2_M3.6 GB4.7 GB8 GB cards
UD-IQ3_XXS4.0 GB5.0 GB8 GB cards
UD-Q2_K_XL4.1 GB5.1 GB8 GB cards
Q3_K_S4.3 GB5.3 GB8 GB cards
Q3_K_M4.7 GB5.7 GB8 GB cards
UD-Q3_K_XL5.1 GB6.1 GB8 GB cards
IQ4_XS5.2 GB6.2 GB8 GB cards
IQ4_NL5.4 GB6.4 GB8 GB cards
Q4_05.4 GB6.4 GB8 GB cards
Q4_K_S5.4 GB6.4 GB8 GB cards
Q4_K_M5.7 GB6.7 GB8 GB cards
Q4_15.8 GB6.9 GB8 GB cards
UD-Q4_K_XL6.0 GB7.0 GB8 GB cards
Q5_K_S6.4 GB7.4 GB8 GB cards
Q5_K_M6.6 GB7.6 GB12 GB cards
UD-Q5_K_XL6.7 GB7.8 GB12 GB cards
Q6_K7.5 GB8.5 GB12 GB cards
UD-Q6_K_XL8.8 GB9.8 GB12 GB cards
Q8_09.5 GB10.5 GB12 GB cards
UD-Q8_K_XL13.0 GB14.0 GB16 GB cards
BF1617.9 GB18.9 GB24 GB cards

Expected speed on popular hardware

HardwareBest quant that fitsExpected tok/s
NVIDIA RTX 3060Q8_032.3 estimate
NVIDIA RTX 4060 TiQ5_K_S38.5 estimate
NVIDIA RTX 3090Q8_068.9 estimate
NVIDIA RTX 4090Q8_076.3 estimate
NVIDIA RTX 5090Q8_0121 estimate
AMD Radeon RX 7900 XTXQ8_076.6 estimate
Apple M4 Pro (20-core GPU) (64 GB)Q8_024.3 estimate
Apple M3 Max (40-core GPU) (64 GB)Q8_035.5 estimate
AMD Ryzen AI Max+ 395 (Radeon 8060S)Q8_021.8 estimate

Check your own card

Too big for your card?

Run Qwen3.5-9B on a rented 24-80 GB GPU by the hour.

Rent a bigger GPU on RunPod

Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.